Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Abstract - Concentration distributions of solid particles transported by pseudoplastic fluids in laminar regime were determined
experimentally. Practically all the distributions show minimum concentration of solids near the centre of the pipe and higher toward the
walls and even, for some conditions, the highest values near the top of the pipe. This behaviour of the concentration distribution was
explained when diffusive fluxes of the solid particles due to concentration gradient and viscosity gradient are considered. An existing
diffusive model was modified in order to take into account negatively buoyant particles and a pseudoplastic fluid as a carrier. It was
found that the head loss plays an important role in the fluxes of solid particles due to gradients of concentration and viscosity, increasing
them enough to overcome the effect of gravity.
HTFF 125-1
pipe with an inner diameter of 5.08 cm. Along the pipe, 3 pressure transducers were installed to get continuous pressure
records. Near the end of the pipe, an electrical resistance tomography (ERT) sensor with a data acquisition system model
P2+ made by ITS was installed in order to record the concentration distribution of solids. The software used for data
processing was ITS System 2000. The slurry returned to the head tank by means of a pipe of 2.54 cm diameter. A refrigerating
system was installed along this pipe that kept the slurry at a constant temperature. Discharge was measured with a magnetic
flowmeter SIEMENS 3100 and a signal transmitter SIEMENS MAG-500.
Head tank
PVC pipe
ERT sensor
Refrigerating system
Flowmeter
2.2. Materials
Glass microspheres of three different size distributions and density (𝜌) equal to 2500 kg/m3 were the solid particles
used in the experiments. Characteristic diameters of each size distribution are given in Table 1.
The pseudoplastic fluid used as a carrier was generated dissolving sodium carboxymethyl cellulose (CMC) in water.
When the solids particles are added to the fluid, the finest particles and the fluid behaves as an equivalent fluid and the
rheology to this mixture needs to be known. Three different concentrations (weight/weight) of CMC in water were
considered: 0.5%, 0.6% and 0.8%, and concentrations (volume/volume) of solids in the CMC and water mixture ranging
from nominal 17% to 32%. Thus, a total of 26 different conditions associated to the materials were defined. The largest
particles of the different types of solids present a process of sedimentation during the rheology determination, and it was
found that particles with sizes smaller than 60-70 m do not sediment through the rheological tests. The characteristic
rheology corresponds to the mixture of fines particles and the pseudoplastic aqueous CMC solution. Rheology was
determined with a rheometer made by Anton Paar model Rheolab QC, with a peltier temperature control. The rheograms
fitted well to a Ostwald–de Waele (or power law) model, 𝜏 = 𝐾𝛾̇ 𝑛 , where 𝜏 is the shear stress, 𝛾̇ is the strain rate, 𝐾 is the
consistency coefficient and 𝑛 is the flow index. Density (fine particles+fluid) was measured with was measured by a Gamma
RTM Dr J Ambrus densimeter. A summary of the parameter characterizing the mixtures and the solid concentrations () are
presented in Table 2.
HTFF 125-2
2.3. Flow discharge
The volumetric discharge 𝑄 of the mixtures ranged between 0.030 and 1.748 L/s depending on each specific the
mixture, and it is given in Table 2.
3. Experimental Results
A total of 241 concentration distributions of the solid particles (for equal number of flow conditions) were obtained
with the electrical resistance tomography sensor. Grouped according to the size 𝑑50 of the particles, the number of
concentration distributions is: 78 for 𝑑50 = 120 m, 81 for 𝑑50 = 300 m and 82 for 𝑑50 = 600 m. Given the impossibility
to show the measurements for all the experiments, 9 were chosen to be presented in this paper and they are shown in Fig. 2.
(The complete set of measurements can be found in the appendix of the Master Thesis of the second author [5], downloadable
from http://repositorio.uchile.cl/handle/2250/132978 ). Associated to the tomographic images presented in Fig. 2 are the
HTFF 125-3
concentration profiles along the vertical axis 𝑧 that are shown in Fig. 3. The value of the concentration for each location 𝑧
corresponds to the horizontal average.
Fig. 2: Concentration distributions obtained with the ETR sensor. Each column from left to right corresponds to d50 = 120 m, 300 m
and 600 m. Each file from bottom to top corresponds to the nominal discharges 𝑄 = 0.3 L/s, 0.8 L/s and 1.4 L/s. Concentration
below 0% does not have physical sense. The value of -10% was chosen to delimit the pipe boundaries.
The measured concentration presented unexpected distributions (at least for the authors), showing lower values near
the centre of the pipe and sometimes with the highest values near the top of the pipe, as can be seen in Fig. 3. After reviewing
the hardware, the data acquisition and data analysis protocols it was discarded that they were measurements artefacts or data
misprocessing, and the results should be analysed considering the dynamics of a non-Newtonian fluid flow and its interaction
with the solid particles.
HTFF 125-4
particles, and it has to be modified for an Ostwal-de Waele fluid and settling particles. According to Phillips et al model, the
transport of concentration is given by:
𝐷
= −∇ ∙ (𝑁𝐶 + 𝑁 ) (1)
𝐷𝑡
where 𝐷⁄𝐷𝑡 is the material derivative of the concentration . 𝑁𝐶 and 𝑁 are the fluxes of particles due to the gradient
of concentration and to the gradient of viscosity, respectively, and they depend on dimensionless diffusion coefficients 𝐷𝐶
and 𝐷 that were determined experimentally by Phillips et al. The fluxes are given by:
In the above equations, 𝑎 is the particle diameter and 𝜇𝑚 the viscosity of the mixture, which depends on the solids
concentration. Note that the flux due to viscosity gradient cannot exist in the flow of a pure Newtonian fluid.
Fig. 3: Distribution of the horizontal averaged concentration. Each column from left to right corresponds to d50 = 120 m, 300 m and
600 m. Each file from bottom to top corresponds to the nominal discharges 𝑄 = 0.3 L/s, 0.8 L/s and 1.4 L/s.
Phillips et al. model does not consider the downward flux resulting from the action of gravity. In the range of Stokes,
this flux is given by
2 𝜌𝑆 − 𝜌
𝑁𝑔 = − 𝑓𝑔𝑎2 (4)
9 𝜇𝑒𝑓𝑓
𝑓 = (1 − )𝛽 (5)
HTFF 125-5
4.8 − 𝛽 2⁄(2−𝑛)
0.57 𝑑 0.27
= 0.0365(𝐶𝐷 𝑅𝑒𝑃 ) [1 − 2.4 ( ) ] (6)
𝛽 − 2.4 𝐷
where 𝑅𝑒𝑃 = 𝜌𝑉 2−𝑛 𝑎𝑛 ⁄𝐾 is the particle Reynolds number (𝑉 is the particle velocity) and 𝐶𝐷 the drag coefficient that
can be computed from any available relationship, like that by Dhole et al. [9] which is valid in the range 5≤ 𝑅𝑒𝑃 ≤500 :
⁄(2.42𝑛+0.918)
𝐶𝐷 = (24⁄𝑅𝑒𝑃 ) (1 + 0.148𝑅𝑒𝑃2.35𝑛 ).
𝑧
In order to carry out a qualitative analysis, the flow in a cylindrical 𝑧
pipe will be simplified to a two dimensional Pouseille flow, where the 𝑧
direction is along the diameters of the pipe (Fig. 4). As gravity acts in the
vertical direction, 𝑁𝑔 will be projected along 𝑧 in the analysis that follows. 𝜃
Assuming a steady state flow with no secondary currents, i.e. only with the
component 𝑢 of the velocity in the 𝑥 direction, the momentum equation is
reduced to 0 = − 𝜕𝑃⁄𝜕𝑥 + 𝜕𝜏𝑧𝑥 ⁄𝜕𝑧, where 𝑧 is the coordinate normal to
𝑥, 𝑃 is the pressure and 𝜏𝑧𝑥 the shear stress that is reduced to 𝜏𝑧𝑥 = Fig. 4: Simplification of the cylindrical geometry
𝐾(𝜕𝑢⁄𝜕𝑧)𝑛 . To simplify the notation, it is defined 𝑃𝑥 = 𝜕𝑃⁄𝜕𝑥 and 𝛾̇ = to the two dimensional Poiseulille flow.
𝜕𝑢⁄𝜕𝑧. Integrating the momentum equation with respect to , 𝛾̇ =
((𝑃𝑥 𝑧 + 𝐶1 )⁄𝐾 )1⁄𝑛 is obtained, with 𝐶1 a constant of integration. The mixture of fine particles and the psudoplastic carrier
behave as an equivalent pseudoplastic fluid, characterized by a mixture consistency coefficient, 𝐾𝑚 , which is a function of
the volumetric concentration of solids, . It can be estimated according to the relationships of Kawase and Ulbrecht (1983)
[10]. Thus, the deformation shear rate of the mixture can be written as
𝑃𝑥 𝑧 + 𝐶1 1⁄𝑛
𝛾̇𝑚 = ( ) (7)
𝐾𝑚
𝑛−1
𝑃𝑥 𝑧 + 𝐶1 𝑛
(8)
𝜇𝑚 = 𝐾𝑚 |𝛾̇𝑚 |𝑛−1 = 𝐾𝑚 | |
𝐾𝑚
Using 𝛾̇𝑚 instead of 𝛾̇ , and 𝜇𝑚 in the expressions for the fluxes, and considering two dimensional Pouseuille flow, it
is possible to get more manageable relationships that will allow us to know the flux of solid particles in the 2D pipe. Thus,
the fluxes associated to concentration gradient and viscosity gradient are reduced to:
𝑑|𝛾̇𝑚 | 𝑑𝜙
𝑁𝐶 = −𝐷𝐶 𝑎2 (𝜙 2 + 𝜙|𝛾̇𝑚 | ) (9)
𝑑𝑧 𝑑𝑧
2
𝑎 𝑑𝜇 𝑚
𝑁𝜇 = −𝐷𝜇 |𝛾̇𝑚 |𝜙 2 ( ) (10)
𝜇𝑚 𝑑𝑧
𝑑𝜇𝑚 𝑑𝐾𝑚 𝑑𝜙
= (11)
𝑑𝑧 𝑑𝜙 𝑑𝑧
With respect to the diffusion coefficients, 𝐷𝐶 = 0.43 [11] and 𝐷 = 𝐷𝐶 ⁄(0.01042 + 0.142) [12] were used. It is
worth to stress the important role played by the energy loss 𝑃𝑥 through the mixture viscosity in all the fluxes. The absolute
value of the shear rate should be used in the equations because the diffusive model is based on the frequency of particle
collisions which scales with |𝛾̇𝑚 |, according to the model of Leighton and Acrivos [7, 8].
It is easy to see that the direction of the fluxes 𝑁𝐶 and 𝑁𝜇 is defined by the sign of 𝑑𝜙⁄𝑑𝑧, 𝑑|𝛾̇𝑚 |⁄𝑑𝑧 and 𝑑𝜇𝑚 ⁄𝑑𝑧.
The flux of particles due to gravity, 𝑁𝑔 , depends only on (𝜌𝑆 − 𝜌) and it is always downwards for negatively buoyant
particles. The net flux of particles along the 𝑧 direction is 𝑁𝑇 = 𝑁𝐶 + 𝑁𝜇 + 𝑁𝑔 sin 𝜃. The result of an analysis for different
conditions is given below for some particular cases.
HTFF 125-6
INCREASING VISCOSITY
Fig. 5: Effect of viscosity in the direction of the flux of particles. 𝑄 ~ 1 L/s and 𝑑50 = 600 m.
Fig. 6: Effect of particle size in the direction of the flux of particles. 𝑄 ~ 1 L/s.
Fig. 7: Effect of the head loss in the direction of the fluxof particles. 𝑄 ~ 1.6 L/s and 𝑑50 = 120 m.
For a relatively high discharge (𝑄 ~ 1 L/s) and 𝑑50 = 600 m, it is found that |𝑁𝑔 sin 𝜃| > |𝑁𝐶 + 𝑁𝜇 | only for the less
viscous mixtures. Radial fluxes (along 𝑧) for this condition are presented in Fig. 5, in which viscosity increases from left to
right. A similar analysis shows that for the same viscosity and discharge, |𝑁𝑔 sin 𝜃| > |𝑁𝐶 + 𝑁𝜇 | in the centre of the pipe.
Fig. 6 corresponds to the cases in which discharge, viscosity and pressure gradient are kept constant, changing the particle
size. It is observed that for the two largest sizes of particles (𝑑50 equal to 300 and 600 m) gravity fluxes dominates only
near the center of the pipe, with strong fluxes towards the walls due to the gradient of concentartion and viscosity. The effect
HTFF 125-7
of the head loss is presented in Fig. 7, where the flux directions are shown for 𝑄 ~ 1.6 L/s, 𝑑50 = 600 m, 𝐾 = 0.29 Pasn,
and 𝑛 = 0.60, and three pressure gradients: 𝑃𝑥 ⁄𝜌𝑚 𝑔 = 0.10 m, 0.25 m, 0.30 m. It is observed that at higher head loss per
unit length, diffusive fluxes overcome the gravitational one.
4. Conclusion
The qualitative analysis of fluxes of solid particles due to gradient of concentration and viscosity indicates that the
concentration of particles carried by a pseudoplastic fluid in laminar regime can present minimum values near the centre and
higher close to the walls. The analysis, although highly simplified, preserves the most important physical mechanisms that
govern the migration of the solid particles. Thus, it was explained why larger particles did not settle in the experiments. It
was found that the pressure gradient (head loss) controls the fluxes through the effective viscosity of the mixture formed by
the solid particles and the pseudoplasic fluid.
Acknowledgements
The authors acknowledge the funding provides by grants Fondecyt 1130910 and Fondequip EQM-120197. The second
author also acknowledges the fellowship PFCHA/Magister Nacional/2013 folio 221320466.
References
[1] A. Oroskar and R. Turian, “The critical velocity in pipeline flow of slurries,” AIChE Journal, vol. 26, no. 4, pp. 550-
558, 1980.
[2] A. P. Poloski, F. Nigl, H. E. Adkins, M. J. Minette, J. Abrefah, J. J. Toth, A. M. Casella, J. M. Tingey, R. E. Hohimer,
and S. T. Yokuda, “Deposition Velocities of Newtonian and Non-Newtonian Slurries in Pipelines,” Pacific Northwest
National Laboratory, PNNL-17639, WTP-RPT-175 Rev. 0, 2009.
[3] P. Mishra and G. Tripathi, “Transition from laminar to turbulent flow of purely viscous non-Newtonian fluids in tubes,”
Chemical Engineering Science, vol. 26, no. 6, pp. 915-921, 1971.
[4] A. Metzner and J. Reed, “Flow of non-Newtonian fluids-correlation of the laminar, transition, and turbulent-flow
regions,” AIChE Journal, vol. 1, no. 4, pp. 434-440, 1955.
[5] G. Montserrat, “Estudio experimental de las pérdidas de carga en tuberías con flujos de mezclas sólidos líquido no-
newtoniano en régimen laminar,” Master Thesis in Water Resources and Environment, Department of Civil
Engineering, University of Chile, 2015.
[6] D. Leighton and A. Acrivos. “The shear-induced migration of particles in concentrated suspensions,” Journal of Fluid
Mechanics, vol. 181, pp. 415-439, 1987.
[7] D. Leighton and A. Acrivos, “Measurement of shear-induced self-diffusion in concentrated suspensions of spheres,”
Journal of Fluid Mechanics, vol. 177, pp. 109-131, 1987.
[8] R. Phillips, R. Brown, R., A. Graham, and J. Abbott. “A constitutive equation for concentrated suspensions that accounts
for shear induced particle migration,” Physics of Fluids, vol. 4, no. 1, pp. 30-40, 1992.
[9] S. D. Dhole, R. P. Chhabra, and V. Eswaran, “Flow of power-law fluids past a sphere at intermediate Reynolds
numbers,” Industrial and Engineering Chemistry Research, vol. 45, no. 13, pp. 4773-4781, 2006.
[10] Y. Kawase and J. J. Ulbrecht, “Rheological properties of suspensions of solid spheres in non-Newtonian fluids,”
Chemical Engineering Communications, vol. 20, no. 3-4, pp. 127-136, 1983.
[11] K. Zhang and A. Acrivos, “Viscous resuspension in fully developed laminar pipe flows,” International journal of
multiphase flow, vol. 20, no. 3, pp. 579-591, 1994.
[12] N. Tetlow, A. L. Graham, M. S. Ingber, S. R. Subia, L. A. Mondy, and S. A. Altobelli, “Particle migration in a Couette
apparatus: Experiment and modeling,” Journal of Rheology, vol. 42, no. 2, pp. 307-327, 1988.
HTFF 125-8